SOLAR POWER GENERATION PREDICTION USING BIGRU MODEL

Abstract

The progress in the field of renewable energy is becoming increasingly crucial as society endeavors to seek sustainable solutions to meet global energy demands. Among various renewable energy sources, solar energy stands out with significant potential to provide clean and efficient power. Forecasting the electricity output from solar power systems has become essential to optimize the utilization of solar energy and efficiently manage power resources and grid operations. In this study, to forecast solar power generation for a plant in central Vietnam, the research team developed a forecasting model known as the Bidirectional Gated Recurrent Unit (BiGRU) with a two-way feedback node. The research focuses on investigating the impact of the number of hidden layers in the BiGRU model on forecasting performance. The research findings indicate that, with a comprehensive dataset, a BiGRU model with a single hidden layer achieves forecasting performance equivalent to that of a BiGRU model with two hidden layers, while significantly reducing computational time. Based on these results, the research team aims to identify the most effective solution for solar power forecasting. Simultaneously, we aim to provide a clear perspective on the benefits of using a BiGRU model with a single hidden layer compared to a BiGRU model with two hidden layers in predicting solar power generation in Vietnam.